Semi-Automated Microscopic Traffic Flow Simulation Development Using Smart City Data

Abstract

Microscopic traffic simulation models have been widely used by transportation planners and engineers for conducting various road network planning and traffic engineering tasks. Due to data limitations, traffic simulation models are often calibrated based on macroscopic traffic measures. Recently, emerging smart city sensor technologies are enabling continuous collection of large volume, high-resolution trajectory data of road users, making it possible to estimate some behavioral parameters of traffic simulation models directly from these data. This research is intended to explore this opportunity with the objective of developing a methodology to estimate traffic simulation model parameters from smart city data with semi-automated calibration procedures. A comprehensive set of calibration procedures are proposed, including both direct methods of estimating parameters from data and indirect methods of estimating some parameters using an optimization algorithm. Most of the proposed procedures are designed in such a way that they can be completed in a semi-automated way using simple Python scripts. The developed methodology is illustrated in a case study involving the calibration of a VISSIM model using an available dataset of vehicle trajectories - NGSIM (Next Generation Simulation) traffic data. While most parameters can be estimated directly from the dataset, some parameters from the selected parameter set are determined using a neutral neural network. The modelling results suggest that the best performing parameter set generates less than 10% error relative to the field measurements in term of travel time and speed.